Messages in this thread | | | Date | Sun, 25 Oct 1998 22:55:29 +0100 (MET) | From | Riccardo Facchetti <> | Subject | Re: ARMS WAVING!!! Proposal to fix /proc dainbrammage. |
| |
On Sun, 25 Oct 1998 robbie@scot-mur.demon.co.uk wrote:
> > What about them? fscanf should be able to handle the above type of > > format quite nicely. > > It's not as efficiant as reading binary. Imho a lot of /proc could be > replaced by user space utilities for humans and shell scripts. The kernel > shouldn't have a ui.
I'll just put my cent. IMHO /proc should be human readable. There is an entire filesystem built to let the users have the /proc files that can be catted etc etc.
Following your reasoning why, for the sake of Crist, we have written an entire filesystem ? What about system calls ? Aren't them much more efficient ?
But obviously having a /proc filesystem is more handy than have a set of system calls (or a single ... a la' pstat()). So what's the point ? We have a filesystem. We have cat'table files. The files _have_ to be human readable. If not, the entire meaning and usefulness of /proc is _shit_. We can get rid of 300 Kb of fs/proc/* and write a new system call in kernel/*.
Again ... I'm not suggesting to rm -rf fs/proc/ ... I just want to point out that the existence of /proc is justified by having some files that can be catted to have some vital informations handy and with the word handy I mean human-readable. Formatting of numbers can be done by the kernel. The routines that do it are called only when 'cat /proc/something' so there's no loss of efficiency for the kernel itself.
And ... what about all that printk about ide-reset ... scsi-timeout ... etc etc ... why not coding all that messages in a single 32 bit number ? So we can get rid of _all_ the kernel human readable messages. No thanks!!!
Blah ... this discussion is getting boring.
Ciao, Riccardo.
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
| |